John M McIntosh wrote:

>
> On Mar 31, 2004, at 2:47 PM, Alan Grimes wrote:
>
>> om,
>>
>> I've been obsessed with hacking the VM for the last 5 days or so,  
>> often staying up till 3AM watching builds compile....
>

> XCode from Apple allows you to distribute builds using GCC 3.3 to all  
> the machines in your house, couldn't say how that is supported
> on Linux or other platforms sorry. This helps immensely if you've 4 
> or  5 macs about the house and you call also steal cycles from your's 
> son's  laptop to boot.


I hadn't really considered that because I can't even get networking to 
work properly on this (linux) machine. I do have a spare athlon 800 with 
6 gigabytes of unpartitioned hard disk space. ( The DOS partitions on 
the drive are indispensable..)

For a dialup internet connection, a 10Mbps lan is excessive. I've moved 
as much as 300/mb in 30 minutes over it.  For a clustering application 
such as this, however, I would have to upgrade both a nic and the hub to 
100mbps at least...

(Linux machine (athlon 800) to linux machine (dual 1.2ghz): 50k/second, 
Linux machine to 486-class DOS machine: 250k/second, go figure.)

It's not really a concern because this is a dual Athlon and it comples 
*FAST*. I can make a VM in about 1:30 minutes. The only gripe I have is 
when I am making the VM sources because it is confined to a single chip. 
=( I like 'round glass' alot but it loves its CPU cycles...

The reason I'm putting VM hacking asside from now is that I am unable to 
get a stable squeak VM working, IE: one that I can run the SUnit tests 
on, save with a possibly broken VM (I only save if it has the same 
failure pattern as my workhorse VM) make changes to the source, and then 
run again ad nausium... I have found that even with the latest 3.6, it 
will crash after doing this several times!  With these problems I have 
no idea wheather I am introducing bugs or making improvments....

If anyone wants to play with the changes I have made so far is welcome 
to e-mail me off list for my current code,

Be warned, it is known to be broken so only try it piece by piece... A 
VM generated directly from it will likley segfault.

Performance wise, I have not been very careful testing but I beleive 
that there are enough "gems" in it to yield a  35-40% improvment in GC 
code and a 1-2% improvment in opcode performance as measured by 
TinyBenchmarks.